24.2 Mode

In the mode tab of the sensitivity analysis the specific sensitivity analysis mode can be selected. The available modi are

24.2.1 Jackknifing
Within the jackknifing mode each file that has been included in index building and different settings are removed iteratively from the index calculation procedure. There are three possible actions:

Figure 70Settings dialog for Jackknifing within the Sensitivity Analysis interface

  • Exluding Files Iteratively: Index A is constructed from the files a, b and c. During the Jackknife procedure with the option Excluding files iteratively three modified index files will be constructed:
    • Modified Index 1 constructed from files b and c (a removed)
    • Modified Index 2 constructed from files a and c (b removed)
    • Modified Index 3 constructed from files a and b (c removed)
  • Excluding Threshold Settings: Index A is constructed from the files a, b and c. With threshold settings for a and c. One modified index file will be created without these threshold settings. If index calucluation did not involve threshold settings, do not select this option.
  • Apply Equal Weighting: The customized weighting scheme that was applied through index calculation will be replaced by an equal weighting scheme and one modified index file will be constructed. Do not choose this option if the weighting of the index already had the character of an equal weighting scheme.
24.2.2 Random Weight Variation

In this mode the weighting scheme that was used for index calculation is varied randomly within defined borders. Define the maximum and minimum weight for each file as the upper and lower border. These should be higher and lower than your original index weight.



Figure 71Settings dialog for random weighting variation within the Sensitivity Analysis interface

Choose between

  • one at a time: the weighting for each file is changed separately
  • all at a time: the weigheiting for all files are changed simultaneaously.

Set the number of files you wish being created during this procedure.

24.2.3 Systematic Weight Variation

In this mode the weighting scheme that was used for index calculation is varied systematically within defined borders. Define the maximum and minimum weight for each file as the upper and lower border. These should be higher and lower than your original index weight. Define the step size as the percentage value that will be iteratively added or subtracted from your upper and lower border.



Figure 72Settings dialog for systematic weighting variation within the Sensitivity Analysis interface

Example: The following settings were made for an index created from file a and b.


Table 1Systematic Weight Variation - Settings
FileWeightMinimum WeightMaximum WeightStep Size
a 30 20 40 5
b 70 50 90 10

Following modified index files will be created from those settings:


Table 2Systematic Weight Variation - Created Files
File 1 2 3 4 5 6 7 8 9 10
file a20253035403030303030
file b70707070705060708090

24.2.4 High and Low Case Scenario
Within this mode the weighting scheme is adjusted according to the distribution patterns of the seperate files. Maximum and a minimum weight has to be assigned for this mode.

Figure 73Settings dialog for high and low case scenario within the Sensitivity Analysis interface

This mode will create two modified index files with modified weightings. The weighting depends on the distribution of the data respectively if the distribution is skewed to the right (lower values are more frequent) or if the distribution is skewed to the left (higher values are more frequent).



Figure 74Graphical explanation of the nonsymetric distribution forms, skewed to the right and skewed to the left

Under the low case scenario minimum weight is given to those files skewed to the left (higher values are more frequent) while maximum weight is given to those files skewed to the right (lower values are more frequent). Under the high case scenario maximum weight is given to those files skewed to the left (higher values are more frequent) and minimum weight is given to those files skewed to the right (lower values are more frequent).